The aim of this work was to achieve two main objectives. The first one was to study the influence of thermal diffusion (Soret effect) on systems containing dissolved solutes in supercritical fluid. Such study consisted in designing and constructing an equipment in which experimental measures for two systems were made, one consisting of a mixture of sunflower oil refined with oleic acid and another of limonene with linalool. These mixtures were used to represent two model systems, one of vegetable oils (I) deacidification and another of orange (II) essential oil deterpenation, respectively. The second objective was to calculate thermodynamic properties using the Peng-Robinson equation of state (EOS-PR) with the classical mixing rule and to estimate the thermal diffusion factor using Kempers model. To that goal, the binary interaction parameters between the pairs of mixture components were obtained by adjusting the thermodynamic model to experimental data (isothermal) of oleic acid solubility, sunflower oil, limonene and linalool in supercritical carbon dioxide. A computational program was developed to calculate thermodynamic properties based on the equations derived from the EOS-PR, properties that are necessary to calculate the thermal diffusion coefficient by the Kempers model. The results showed that the designed equipment was able, with some limitations, to provide experimental results for the two model systems. The collected data showed that thermal diffusion influenced the separation of the mixture components of the model II system, while for model I system there was no fractionation. In the thermodynamic modeling of the binary systems phase equilibrium using the EOS-PR, a good correlation was obtained for the two model systems. However, for model II system, the results showed that the critical properties have significant influence over the thermodynamic modeling performance, and a correction of these values is necessary. For the non-isothermal system modeling, most of the effort to estimate the Soret effect was put in calculating the thermodynamic mixing properties. The obtained values for the thermal diffusion factor showed a strong dependence on the binary interaction parameters and a sensitivity to errors in the mixture thermodynamic properties. In spite of that, the modeling proposed in this work represents qualitatively the experimental data behavior, being able to clarify in which conditions of pressure, temperature and composition the best separation can be obtained. Therefore, the results were promising and will serve as a basis for future developments, since further research in this area is necessary to extend the application of thermal diffusion in supercritical environment to other mixtures to apply modifications in the experimental system and to implement new models to calculate the thermal diffusion coefficient.

pela sua valiosa orientação, pelas oportunidades como PED, pelos ensinamentos, aconselhamentos e apoio, essenciais para meu desenvolvimento pessoal e acadêmico, e pela confiança depositada em mim. Á minha co-orientadora Dra. Losiane Cristina Paviani, pela atenção e dedicação no laboratório e por aperfeiçoar o meu trabalho com sua experiência. Ao CPQBA -UNICAMP, especialmente ao Dr. Pedro Melillo de Magalhães pelo fornecimento da matéria-prima e a Dra. Carmen Lucia Queiroga pela cooperação na etapa experimental do desenvolvimento deste trabalho. Aos professores membros da banca examinadora pela disponibilidade e atenção dispensada ao trabalho, bem como pelas valiosas contribuições. Aos meus professores da Universidade Estadual do Centro-Oeste (UNICENTRO), por todos os inestimáveis ensinamentos e por despertarem em mim o amor pela ciência. Á técnica Patrícia Tonon de Souza, pelo auxílio no laboratório e a todos que compõem o grupo EXTRAE. Em especial, á Julia Teixeira de Paula pela ajuda com o extrator e pelas injeções de otimismo. Ao Fábio pela sua constante colaboração, dedicação e compreensão no decorrer deste trabalho. Obrigada pela paciência e companheirismo. Aos meus pais, Nilo e Ivete Garmus, pela dedicação, sacrifícios e ensinamentos. Á Ana Lúcia e Carlos Eduardo Diniz, por me acolherem como uma filha em Campinas. Aos meus colegas e amigos do DEA, pela amizade e parceria. Ao CNPq e a FAPESP pelas bolsas de estudo concedidas, importantes para a minha formação profissional. A todos que direta ou indiretamente contribuíram para a realização deste trabalho. x xi SUMÁRIO LISTA DE TABELAS .